Re: subroutine, chaning links

[Tim Mooney <mooney@aps.anl.gov>]

Ben Franksen wrote:
> On Freitag, 30. Mai 2008, Mathias Steiner wrote:
> > All would be well if it were possible to change links -- either .FLNK or
> > .OUT -- from within a subroutine.  I've tried a couple of things without
> > success.  Just pasting the channel name into the "value" field of the
> > union does not, and probably should not, work.
> >
> > Assigning a different value to a link by copying an input link doesn't
> > work either:
> > psub->flnk = psub->inpb;
> >
> > This isn't so clear to me, since they are both of type DBLINK.  After
> > all, I can "just" type the value into the .FLNK field of any channel
> > using edm, and it'll work...
>
> It works because this operation is going via Channel Access.
>
> You can do the same from your subroutine: use the CA client library to 
> establish a channel to the FLNK (or whatever other link) field, then do a 
> ca_put with the new value. A very simple example for how to use CA (client) 
> is provides as a makeBaseApp template, just do
>
>   makeBaseApp.pl -t caClient
>
> and look at caExample.c.
>
> I am not sure whether the ca_context_create() call is necessary if you call 
> CA from a subroutine record, maybe the context has already been 
> initialized?
>
> Cheers
> Ben

I think I would not do this from within a sub record's subroutine, even
though it's possible.  I would probably arrange for an SNL program, or
for some string record's output link (with attribute 'CA'), to write to
the links that need to be changed.  A sub record's subroutine is not well
positioned to perform as a CA client, and there seems to me little benefit
to be gained, in the end, from coercing this.  The subroutine can't use the
new links until it returns and is re-entered (i.e., until the lock set has
been released, and CA has had an opportunity to run), so you'll have the
principal disadvantage of using an explicit third party, whether you
actually use one or not.

--
Tim Mooney (mooney@aps.anl.gov) (630)252-5417
Beamline Controls & Data Acquisition Group
Advanced Photon Source, Argonne National Lab.